Messages in this thread | | | Date | Thu, 08 Jan 2009 20:45:06 -0800 | From | Arjan van de Ven <> | Subject | Re: [PATCH] async: Don't call async_synchronize_full_special() while holding sb_lock |
| |
Dave Chinner wrote:
> > So, given the potential impact of this change, what testing have > you done in terms of: > > - performance impact
I tested this on my machines and it gave a real performance improvement (11 to 8 seconds for a full kernel tree unlink, and cutting out latency for normal applications) > - sync() safety that was exactly the synchronization point that's discussed here.
> - removing a million files and queuing all of the > deletes in the async queues....
the async code throttles at 32k outstanding. Yes 32K is arbitrary, but if you delete a million files fast, all but the first few thousand are synchronous.
| |